Bad scan mirror specs?

[The_Doctor]

Cambridge Technology have a mirror specified for 3mm aperture, with a PDF containing a diagram of it here:
http://www.camtech.com/images/produc...#37;20diag.pdf

3mm is the APERTURE, not the actual width. That's 0.2", or 5.08 mm.

They also have a diagram of the galvo mount:
http://www.camtech.com/images/produc...%20D03793B.pdf
Despite the first galvo being canted forward to a 15° angle, the beam path makes it clear that this is an orthogonal layout, that angle slightly helps the second mirror aperture, but does nothing for the first.

Scan angles, beam widths, mirror widths and apertures can't be specified without specifying all four values at once. Different combinations of values exist for the same physical system, but you still have to specify all four.

Look at what is claimed. 3mm aperture from the 5.08mm mirror. Assuming that beam width equals aperture to get the best possible scan for max beam width for the 3mm aperture and to simplify the calculations, this equal value must be no more than 5.08 * COS(45°+X) where X is half the mechanical scan angle. This equates to 8.75° for X, or 35° optical, which is some way short of the claimed 40° that I thought these systems were specified for.

Isn't this bad specsmanship? Isn't it the same (though I grant, not as bad by degree) as selling an ion laser that can do 4 watts, or run for 2000 hours, without telling the end user that it can't do both at once? While a thin beam can scan at 40° from a 5(.08)mm mirror, trying to get a 3mm beam to scan 40° will lose light. We can try to optimise angles and positions, but unless the beam profile is elliptical it won't make it efficiently past the first mirror, and certainly won't make it intact past the second.

[Pangolin]

Hi Doc,

Just a few points. First, you don't really have to specify "all four". All you only need to specify is the "cone angle" (actually a pyramid) and the associated "clear aperture" that can pass through and by projected by that cone angle.

In the case of the X mirror, the most important constraint is its width. It must be sufficiently wide so that it can "catch" the entire beam through all angles. If you look at Cambridge scanners, you will notice that the X mirror is wider than the Y mirror. This is necessary because the beam enters the X-Y system "horizontally".

The 5.08 mm width of the Y mirror is able to reflect a 3mm beam upward by 30 degrees optical and downwards by 30 degrees optical. The "upwards" is of course the worst case for the Y mirror because that is the point at which it appears the "thinnest" to the entrance beam. Note that if the entrance beam was vertical and not “canted by 15 degrees”, the Y mirror would have to be much wider.

"Canting" the X scanner back by 15 degrees is a trick that can be done to reduce the width of the Y mirror. This "cant" can be any angle (we used up to 45 degrees years ago), but eventually the beam exiting the Y mirror will actually be re-reflected by the X-mirror. An angle of 15-degrees is a good all-around number and works well for 60-degree cone angle. In current X-Y mounts, we use this number even for smaller angles but our reason for doing this has to do with mirror inertia, not scan angle.

Incidentally, if you configure an external mirror system (even just a single mirror), and enter an X-Y mount not horizontally, but at a 15-degree downward angle, you can reduce the width of the X mirror as well. The resulting system would still be considered an "orthogonal" mounting arrangement.

It could be that you are confused because you have latched onto the number "80-degrees" (40 degrees mechanical), as many people do. Although the scanner itself can scan at up to 80-degrees (actually more), it is MOST OFTEN that the servo system, X-Y mount, and mirrors on the scanner are configured to scan "only" as wide as 60-degrees.

Also, one more point is worth mentioning. Chinese companies use a Y mirror for both the X scanner and Y scanner. They must think this is a smart thing to do because a) the inertia is the same and b) fewer individual parts to inventory. But it isn't a smart thing to do (for the client). Since the X axis doesn't have this "cant" angle, it means that the clear aperture for Chinese scanners is NOT 3mm when projecting imagery up to 60 degrees wide. Fortunately, most Chinese scanners are used with Chinese lasers with a beam diameter smaller than 3mm…

Best regards,

William Benner

PS: The mirror sizes on Cambridge scanners were designed back in 1992 and, aside from the thickness, they haven’t changed. Of course, all of this is done by CAD these days, and usually some advanced program like SolidWorks or Inventor. In any event, over the 15-year period that this mirror size has been used, it seems likely that if there was a mistake made in these mirror sizes, it would have been caught by now

[The_Doctor]

Humungous post alert.
Please read carefully before thinking I missed something. I might have, but I did try not to.

Hi Doc,

Just a few points. First, you don't really have to specify "all four". All you only need to specify is the "cone angle" (actually a pyramid) and the associated "clear aperture" that can pass through and by projected by that cone angle.

I originally said three not four, and edited... I still think you need three: mirror width, scan/cone angle, aperture. That's because there are other ways to arrange the beam path and any cant angles in the system. You can reduce to scan angle and aperture for a complete system with fixed mount and mirrors, but when designing the system you need a bit more.

In the case of the X mirror, the most important constraint is its width. It must be sufficiently wide so that it can "catch" the entire beam through all angles. If you look at Cambridge scanners, you will notice that the X mirror is wider than the Y mirror. This is necessary because the beam enters the X-Y system "horizontally".

I don't have access to any to look at directly, but I noticed the 0.236 in the diagram. I thought the extra beyond 0.2 was just for the mount to grip, the diagram doesn't make this clear. Looking at it again (taking 0.236" as 5.99mm and rounding up to 6mm, and assuming +/- 15° mechanical scan angle), 6 * COS(45°+15) = 3mm. An exact fit (to within 11 decimal places!). Nice.

The 5.08 mm width of the Y mirror is able to reflect a 3mm beam upward by 30 degrees optical and downwards by 30 degrees optical. The "upwards" is of course the worst case for the Y mirror because that is the point at which it appears the "thinnest" to the entrance beam. Note that if the entrance beam was vertical and not “canted by 15 degrees”, the Y mirror would have to be much wider.

None of that was lost on me, but I hadn't calculated the scan angle for 15°, I'd only found this mount yesterday while looking for some details. I just checked it and it works ok.

"Canting" the X scanner back by 15 degrees is a trick that can be done to reduce the width of the Y mirror. This "cant" can be any angle (we used up to 45 degrees years ago), but eventually the beam exiting the Y mirror will actually be re-reflected by the X-mirror. An angle of 15-degrees is a good all-around number and works well for 60-degree cone angle. In current X-Y mounts, we use this number even for smaller angles but our reason for doing this has to do with mirror inertia, not scan angle.

I noticed. The X mirror being wider can be a closer match to the inertia in the Y mirror. My idea (turns out not to be patentable, which removes that headache) was to make the cant angle 30° for a 60° beam path through the system, for both mirrors. I hadn't looked for optimisations for mirror inertia but I had proved that a 3mm width for both mirrors can scan a beam greater than 2 mm wide through 45° optical scan angle without significant obscuration of the final output by a corner of the (bevelled) X mirror. I'd also got the mirror axis separation greater than with the CT mounts. This is important with WideMoves and possibly other scanners where the mirrors could collide during startup/shutdown transients because they park in arbitrary positions. I had wanted to make the X mirror wider to take advantage of the extra space available, but unless I could also shorten the Y mirror a bit to speed it up and get their inertias to match better, there didn't seem to be much point. There is also an expense involved in overspecifying exact mirror dimensions with any complexity.

Incidentally, if you configure an external mirror system (even just a single mirror), and enter an X-Y mount not horizontally, but at a 15-degree downward angle, you can reduce the width of the X mirror as well. The resulting system would still be considered an "orthogonal" mounting arrangement.

Personally I think that's a loose use of word 'orthogonal' but having looked at how it is now conventionally used, it would be splitting hairs to disagree, the output has linear perpendicular axes so that's still orthogonal. I knew the extra input angle change was useful, my idea depended on it. My idea is still very useful. It can really help with WideMoves and wider beams from cheap lasers, all you need is to make a 30° wedge to lift the X mount off the base plate. This is well within the reach of anyone with a decent bandsaw. As a lot of fast graphic scans and beam shows are aimed at getting high speeds at narrower optical angles, my idea of a 60° beam path would be good for the high end of the market too, as a standard. Over 2.1mm beam width at 45° optical off mirrors only 3mm wide, placed such that they cannot possibly collide? I bet there'd be a few takers for that.

It could be that you are confused because you have latched onto the number "80-degrees" (40 degrees mechanical), as many people do. Although the scanner itself can scan at up to 80-degrees (actually more), it is MOST OFTEN that the servo system, X-Y mount, and mirrors on the scanner are configured to scan "only" as wide as 60-degrees.

Not a chance... I might have made various mistakes but not that one. I had made the assumption that most narrow-angled fast systems would only be 40° optical angle though, so with my design I decided to go to 45° and see how good it could get. I tried various cant angles. I decided 30° (for 60° beam path) was optimal, as it's the easiest to make with modest tooling, and gets excellent results.

Also, one more point is worth mentioning. Chinese companies use a Y mirror for both the X scanner and Y scanner. They must think this is a smart thing to do because a) the inertia is the same and b) fewer individual parts to inventory. But it isn't a smart thing to do (for the client). Since the X axis doesn't have this "cant" angle, it means that the clear aperture for Chinese scanners is NOT 3mm when projecting imagery up to 60 degrees wide. Fortunately, most Chinese scanners are used with Chinese lasers with a beam diameter smaller than 3mm…

German too. WideMoves do it, with a 0.5 mm thick mirror on a tiny spindle. When I claimed I saw problems with those I ended up jumping through hoops for DAYS on alt.lasers to prove my case! Given what I've learned since about parasitic resonance (already suspected even then in addition to the hysteresis problem I already identified), I'm surprised that no-one pointed out that not only would I have these problems, but it could hardly have been otherwise. The lack of cant angle in most cheap systems was what led me to work out my idea. I'd not seen any CT scanners before either, and had assumed this basic 90° plan was standard. No-one had ever told me otherwise. So, ok, I can't patent. But I don't care. The best way to understand a wheel is to re-invent it thoroughly.

I propose that 2mm aperture systems be built with 3mm wide mirrors for extra speed at smaller angles with the cant angle doubled from 15° to 30° and a greater separation between mirror axes to accommodate that and to prevent possibility of mirror collision. A slight widening of the X mirror and a reduction from 45° to 40° optical angle should allow a sightly shorter distance between mirror axes, and a significant shortening of the Y mirror. Their shapes will differ but their inertias will be more similar as seen by their drive motors. If anyone is interested in this, I'll adapt my design so it can be used as an easy basis for getting measures for machining parts, and I'll post it here for anyone to use. As I mentioned, WideMove users already have a cheap and easy way to explore this 60° beam path if they have beams up to 4.5 mm wide.

A last (important) point is that although the CT design allows a horizontal input, this solves nothing. If a laser and a galvo both need good heatsinking, a mismatch in height means compromising thermal coupling while making shims/blocks to adjust to a matching relative height on the same baseplate. With my idea you have to use a mirror to steer the beam into the scanner block, but this isn't as bad as it sounds. First, you only have to rotate the laser a tiny bit on the baseplate to get the beam passing across the axis of the input path. Second, the mirror angle to steer it is shallow so that very good reflection is possible even with a cheap metallised mirror, and is the only part that needs adjusting to optimise coupling between laser and first mirror. If you're wanting to get the galvo block close to the case aperture to minimise window size, you'll have to use a steering mirror anyway, so this single adjustable mirror solves that problem too.

Best regards,

William Benner

PS: The mirror sizes on Cambridge scanners were designed back in 1992 and, aside from the thickness, they haven’t changed. Of course, all of this is done by CAD these days, and usually some advanced program like SolidWorks or Inventor. In any event, over the 15-year period that this mirror size has been used, it seems likely that if there was a mistake made in these mirror sizes, it would have been caught by now

Thickness doesn't matter much to scan angle, I realised. It amounts to an offset in placement, and doesn't affect scan angle much. It actually contributes to the increase of scan angle for a given aperture in my design, by helping the lowest horizontal scan to clear the X mirror. I need to look into Solidworks, or Inventor (which I'd never heard of till you said it there). SketchUp will do it for free, but it won't model motion. It really sucks having to move everything by hand, making sure I didn't miss something. Takes days, when it could take hours or minutes to explore stuff deeply (but it's still more than Euclid ever had to help him). Even when mistakes don't exist, there are usually other ways to do something. Some of them might be better, especially in circumstances such as those that led me to my own results.

[yaddatrance]

I don't have access to any to look at directly, but I noticed the 0.236 in the diagram. I thought the extra beyond 0.2 was just for the mount to grip, the diagram doesn't make this clear. Looking at it again (taking 0.236" as 5.99mm and rounding up to 6mm, and assuming +/- 15° mechanical scan angle), 6 * COS(45°+15) = 3mm. An exact fit (to within 11 decimal places!). Nice.

Some interesting ideas, I haven't had time to think through it, but thought I'd add, the X mirror is actually supposed to be mounted on the wide side.

[The_Doctor]

Yes. I don't know why I didn't consider it before, but I'm in good company, the person who drafted that diagram didn't specify it either.

A correction to a bit of the end of my post, about mirror thickness: It doesn't help the clearance, that's entirely due to the angling of the mirror surface, extra thickness actually makes it a bit more awkward. Even so, the effect is small if the thickness is a small proportion of mirror width and of the axial separation between mirrors. I didn't work out maths for it, I just modelled the way it works and looked for arrangements that worked ok.

[Pangolin]

I still think you need three: mirror width, scan/cone angle, aperture. That's because there are other ways to arrange the beam path and any cant angles in the system.

Well I guess "the number of parameters needed" depends on who is supplying what. Cambridge and other scanner manufacturers supply a complete package -- scanners, X-Y mount, servos, etc. This is a package -- you put a beam in, and you get scanning out. So, for this complete package, what does the user need to know?

Most ordinary human beings can't figure out mirror width and length and, especially, thickness. Heck, even many Chinese manufacturers get this wrong!! So Cambridge and other manufacturers can help end-users by telling them what they really need to know. You put this size beam in, and you can scan that wide, period. End-users who are applying the package don't need to know anything more. Who cares how wide the mirrors are, as long as it works…

And trust me, speaking as someone who deals with end-users all the time, the less you need to tell them in order to get their job done, the better!!

I noticed. The X mirror being wider can be a closer match to the inertia in the Y mirror.

Yes. This is something that Chinese scanner manufacturers miss, and it is something that does not occur to most people. Ideally, you want the mirror inertia to be the same for X and Y, since we are doing vector scanning. Chinese simply accomplish this by using the same mirror for both, but this tremendously sacrifices X scan angle, especially if you want the full 3mm aperture...

My idea (turns out not to be patentable, which removes that headache) was to make the cant angle 30° for a 60° beam path through the system, for both mirrors.

Any angle can be used, as long as the exiting beam does not scan back onto the scanning mirrors themselves. As I said, back in the old G120 days, we routinely used 45 degrees. And as I said, you can save some X mirror width by adding one more stationary mirror to the system, and entering the X-Y mount, not horizontally, but at a downward angle.

Incidentally, this is an example of something that is "known to those skilled in the art". Of course, I bet you would be hard pressed to find a whole lot of people who know these things...

I'd also got the mirror axis separation greater than with the CT mounts. This is important with WideMoves and possibly other scanners where the mirrors could collide during startup/shutdown transients because they park in arbitrary positions.

The Cambridge X-Y mounts for Lightshow folks, is designed such that even if you rotate both scanners so that the mirrors are in the closest proximity, they still won't crash. There is a small gap between them.

This is only done for lightshow folks. For bigger aperture scanners, the mirrors absolutely will crash if you try this. BUT, the crash is prevented by mechanical stop systems, most often found on the front of the scanner.

I had wanted to make the X mirror wider to take advantage of the extra space available, but unless I could also shorten the Y mirror a bit to speed it up and get their inertias to match better, there didn't seem to be much point. There is also an expense involved in overspecifying exact mirror dimensions with any complexity.

Maximum power is transferred when the mirror is the same inertia as the rotor. In the case of these small lightshow scanners, this is not done. The mirror is 1/10 the inertia of the rotor, so the motor portion of a scanner is expending most of its power moving... itself!

My point is, you really can't speed things up by messing with the mirror size, with these small mirrors being what they are, and rotor inertia being what it is. This is a fallacy among Chinese manufacturers who think they can make the scanners go faster by making the mirror thinner. Sorry, they can't... In fact, because of a loss of dynamic stiffness, thinner mirrors actually make the scanning system as a whole go slower.

all you need is to make a 30° wedge to lift the X mount off the base plate.

As I said, back in the bad old days of G120s, we used 45 degrees. But this must be evaluated very carefully. Eventually, with increases of the Y scan angle, the beam that exits the Y mirror will re-enter the X mirror.

German too. WideMoves do it, with a 0.5 mm thick mirror on a tiny spindle.

Having evaluated the WideMoves myself, my personal opinion is that they are not very good. At some point soon I will find time to post a picture or two of our scanner workbench, and everyone will see that I am qualified to make such an assessment…

The lack of cant angle in most cheap systems was what led me to work out my idea.

Unfortunately, there are a whole lot of people in the world who don't understand that scanning is a business of "details". People might see a Cambridge mount and not understand what you call the "cant angle" (in scannerdom, we call it a set-back). Plus, to make the setback requires extra skill on the part of the machinist.

It comes to me as no surprise that very cheep systems don't integrate a set-back. Probably the vast majority of companies don't understand it, and the remaining ones can't implement it.

Even when given a great example of an X-Y mount (Cambridge will give you drawings to make your own for free!), people still don't get it...

Another point of contention is HOW the scanners are secured within the mount. Because these scanners develop heat when they are scanning, a correct mount will be a "compression" style mount, where there is contact all the way around the body of the scanner. A really bad example of an X-Y mount was supplied by NML for years. Not only was their no set-back, but the scanners were secured into the mount by a single screw. If you "do the math", you will see that you only have two points of contact for heat dissipation if you use a screw.

So we have examples of great X-Y mounts (those provided by Cambridge) and we have examples of really bad X-Y mounts (those formerly supplied by NML)...

I propose that 2mm aperture systems be built with 3mm wide mirrors for extra speed at smaller angles with the cant angle doubled from 15° to 30° and a greater separation between mirror axes to accommodate that and to prevent possibility of mirror collision.

Well, a few things. First, the mirrors won't crash with Cambridge mounts. Second, even if you make the mirror 1mm by 1mm and put it on to of the scanner, the scanner will not scan any faster because even with mirrors that are 5mm wide and 8mm long and over 1.5mm thick, the mirror represents only 1/10 the inertia of the scanner's rotor. Even if you completely remove the mirror, the scanner won't go any faster.

A last (important) point is that although the CT design allows a horizontal input, this solves nothing.

As I already wrote in several posts, you don't necessarily need to enter horizontally. You can add a single mirror and enter at a downward angle. As I understand it, you already propose to add a mirror, right?

I need to look into Solidworks, or Inventor (which I'd never heard of till you said it there).

It's basically the way things are done these days. In the bad old days, you would use something like AutoCAD to come up with a design, manually generate the views, then manually generate and document tolerances. Nowadays you just make the part in SolidWorks (what we use) or Inventor. These products automatically make the views and generate the documentation for tolerancing. But this is rarely used to generate prints anyway... Nowadays you just take the SolidWorks file down to the machine shop, they plug the disk into their CNC machine and it makes a perfect part. No more relying on humans to interpret drawings...


Best regards,

William Benner

[The_Doctor]

This is a package -- you put a beam in, and you get scanning out. So, for this complete package, what does the user need to know?

Well, I mentioned that if it's a complete system, I agree, you can reduce the need for info. I think end users like a minimum to get them up and running safely and easily, but once that's started they'll probably look deeper at info that is given them. I know I do, and I'm an end user too.

You put this size beam in, and you can scan that wide, period.

Yes, but what size? This is why I edited the subset to four variables, separating beam width from aperture. Even if you reduce the need for an end user to consider mirror width and separation, you still have the aperture and scan angle, but how much angle do you get for a thinner beam? We know it varies in some proportion to beam width, but an at-a-glance guide in specs would help an end user. MediaLas quote the various expectations of scan angle for a given speed in three or four likely end-user choices, and I think similar small specs tables for beam width versus scan angle are important. Designers can't easily calculate this, so they can't expect end users to do it, especially if empirical efforts to find out might threaten the safe operation of the scanners. I know that max safe scan angles are given, but an end user struggling with table alignments might make an error they'd easily avoid, with just that bit of extra help from the designers. It's like those little tables of speeds and drill sizes inside the lid of a pillar drill. It's pure end-user level of detail, but it's enough to quickly let them get good results in various ways.

Of course, I bet you would be hard pressed to find a whole lot of people who know these things...

Well, yes, which is why I found myself re-inventing the entire X/Y scanner system... I know you said that the downward aiming and angled mirror mounting help, but the fact that I had already re-invented it and chosen a sensible value of 30° proves that it can be done by someone who is not connected to the existing knowledge. This is all deducible. It's easier than trying to find prior art in the patent system, as it happens.... Btw, I also tried it at 45°. Given the narrow originally specified angles for GS-120's, that might have been a wise choice, but by the time I first heard of the incorrect pure-90° forms, I'd also heard of CT6800HP's and WideMoves, so 45° was too restricting. I know any angle will do, and I even contemplated whether the golden mean was involved in an optimum choice, but 30° still seems right to me. I think 15° is good, very good given that the first 15° makes more help than the last, but it's a kind of please-everyone kind of a choice. Surely if you can get 2mm beams to scan 40° off a Y mirror just 3mm wide and 6 mm long (active length), and speed up the system dramatically by doing so, it becomes a viable idea specifically for this.

The Cambridge X-Y mounts for Lightshow folks, is designed such that even if you rotate both scanners so that the mirrors are in the closest proximity, they still won't crash. There is a small gap between them.

Where? I got my info from http://www.camtech.com/images/produc...%20D03793B.pdf where the centre image gives info for the calculation ((0.271-0.222)**2+(0.879-0.714)**2)**0.5. This gives a separation of 0.172" between mirror axis centres, but the mirrors want 0.221" from midpoint to midpoint if they meet at corners. I'd say that 0.254" would be safe, but nothing like this is specified in that diagram, so those mirrors would collide. That diagram was the one they said was specifically for the 3mm aperture mirror set.

Even when given a great example of an X-Y mount (Cambridge will give you drawings to make your own for free!), people still don't get it...

I did. I got it before I ever saw that mirror mount, I got the principle so well I had to momentarily agonise over whether to finish my work first, or halt to go hunting to see if it had been patented already. Ok, so it's been done, but that can't detract from the fact that I can independently work it out, purely from basic observations with a tool not best suited to the task. WideMove users already have it good, given that the 30° wedge and one mirror is all you need to add. Given that they use the same mirror type for X and Y, this is important for them. Incidentally, I bought my WideMoves over two years ago, because I knew they'd be easy to adapt to the 60° beam path if I wanted to try it. In the end I opted for an easier life because I began to realise that WideMoves have more problems than this optimisation could solve.

Even if you completely remove the mirror, the scanner won't go any faster.

For those galvos, perhaps. Nature shows that the creatures with the fastest accelerations of their mass are the smallest. Look at insect legs and wings... This means that even if what I said isn't true for the current scanners, it IS true for those that will eventually be developed.

As I already wrote in several posts, you don't necessarily need to enter horizontally.

And as I was stressing in a similar number of posts, I was already sure than you don't need to. It isn't the best way. I gave several reasons for this, based on the context, where people usually find laser aperture heights that don't conveniently match the galvo block. It didn't escape my notice that neither the CT galvo block or the lasers themselves offer any compromise, so it has to come from somewhere. My design automatically provides a neat answer, as well as allowing easier placement of galvos close to the case edge so the window can be small even if the lasers are inconveniently large. Not a bad set of answers from one tiny mirror, I think.

Question:
Has much work been done on the idea of two drivers operating one gimballed mirror for an X/Y scan? I never gave that much thought because my efforts were clumsy and not worth trying to build, but I can't help thinking that there might be a way to do it. I wondered if it had been done effectively.


EDIT:
Bill, you agreed that it is desirable to get the inertias of each mirror as equal as possible, so isn't this also saying that the difference in mass is significant between the two sizes? If so, then it follows that the reduction to 3mm widths must also be significant, even with current scanners. Either a change between the current Y mirror size and a smaller size is significant, or it is not. If it is, then I am right, If it is not, the Chinese scanner makers are right.

[yaddatrance]

While it looks like the mirrors would collide if they could rotate 360 degrees, they physically can't. There's a cat's whiskers between the two at their closest points.



I've manually canted them so they are at their closest point. In operation
you won't get anywhere this close because of the mechanical stops.

And in terms of lighter mirrors, I remember some noise a couple years back on
CFRFQ "composite mirrors"... It's the sort of problem where I'd imagine if you threw money at it, it'd go away...

Keep plugging at it doc, it's interesting stuff...

[The_Doctor]

Plug.. plug.

I will, I'm after trying to bring the arcane art into the open so we can all see improvements. Re the separation, I wasn't actually trying to suggest that CT galvo blocks don't get it right in this case, no matter how it looks as I write it. I'm aware that a small CT scanner (or a WideMove) can be set to any position manually or by a throw if the power drops out during a fast turn, as they don't use torsion bars. I had no doubt that CT had sorted this, it's just that the specs I saw don't describe a reality that does sort it. I started this thread here mainly because I wanted to explore what appeared to be discrepancies either between published specs and the reality, or between the reality and what might also be acheived.

My guess is that the diagram I saw is older than the last update to the design for units actually sold. It should be updated though, because if people go to the source to learn, whose fault is it if they then get it wrong? Not everyone has the time to second-guess or proofread documentation of a major manufacturer.

Bill, I've thought some more on that mirror mass thing. The force conversion from electric current doesn't see a motor mass and a mirror mass, so I don't understand why you speak of power coupling efficiency on those terms. All it sees is A mass. It can't see the difference if there is enough rigidity there. I agree for this reason that mirrors should be stiff, but I think they should also be light. Smaller is the easiest way to do that, so there should be cant angle adaptations made for systems meant for faster narrower scans. This is no less relevant that using NdFeB magnets for greater flux density despite their lower heat tolerance, or using ceramic rotor formers, or any other optimisation designed to increase force to mass ratio. (Moment of inertia actually, if I remember right.)

Another point of interest, you mentioned the poor thermal contact with a single screw to hold the scanner in a mounting block. I totally agree, it's a problem I solved in a laser diode mount, where I had to have central alignment AND excellent thermal coupling. That CT block is actually a compromise, if it wasn't drilled with high accuracy to match the galvo body, it wouldn't be much better. If the bore was too big the points of contact would be narrow, on each side. If it was too small the contacts would be top and bottom, and although the clamping would more easily increase the area, that same clamp's leverage could easily deform the galvo body, possibly destructively if the design couldn't accept some of that pressure safely.

It's not an easy problem to solve given the large volume of metal there. I'd still try to do it by making the compression work by rotation though, not translation (like a vice). The CT mounts almost seem to do it but with some vice-like action, and I saw one of the Chinese mounts which actually does do it right, though they don't seem to have so much metal to shift so it's easier for them. Looking at Yadda's first pic, I see that there is space to have a bit more metal on the top of the X clamp. If the cut went only into the bore, and not through the bottom of it as well, the clamp would evenly apply pressure on all possible points equally so long as both bore and scanner body were accurately round. I'm sure that CT tried this, so my guess is they decided it was too difficult to stretch that bulk of metal so they followed through with the cuts to make it easier to do.

Edit:
Actually I'd make the side walls thinner. Not much, just enough to allow rotation around from side to top. What thermal conductivity is lost would be more than made up by the increase in metal contact on the bottom. I can't see from that pic whether there is a fine saw cut separating the X and Y clamps. but I'd add one if not, at least down to a little lower than the midpoint level in the X bore.

More edit:
Yadda, how much does a pair of scanners and mounting block like that cost? (With or without amps). Whatever suggestions for improvement or alternative I might have, that's still the best I've ever seen, and I want some.

[yaddatrance]

I think the mounts by themselves are actually ~$100 new from CT as they are reamed to spec (and not just drilled). The set cost ~$1200 complete, and I've personally never bought the mounts by itself... Though I have seen real deals on the mounts by itself on ebay.

The CT's are very well optimized from amp to scanner, but as they use very conventional and inexpensive components. I would be of the opinion that there is plenty of room for improvement, In our current crazy real world you'll find 6' nitrogen cooled mirrors which move faster than you can see.